botfuel-module-collect-tool v0.1.2

Botfuel module collect tool

This module allow to simulate user inputs using a CSV input file that will be processed and generate an output file with the different input transformation steps.

Install

npm i -D botfuel-module-collect-tool

How to use

Configuration

To make it work your configuration file must:

• include the module
• use the adapter provided in the module
• use the nlu module provided in the module
• provide the path of the CSV input file

// file collect-config.js
module.exports = {
  adapter: {
    name: 'collect',
    files: {
      input: './input.csv',
      output: './output.json', // optional
    },
  },
  nlu: {
    name: 'botfuel-collect',
  },
  modules: ['botfuel-module-collect-tool'],
  logger: 'info',
};

Run

The module come with it's own start command botfuel-collect-run, to use it you can create a new script in your package.json like the following:

// file package.json
{
  ...
  "scripts": {
    ...
    "start:collect": "BOTFUEL_APP_TOKEN=<the BOTFUEL_APP_TOKEN> BOTFUEL_APP_ID=<the BOTFUEL_APP_ID> BOTFUEL_APP_KEY=<the BOTFUEL_APP_KEY> botfuel-collect-run collect-config",
    ...
  },
  ...
}

and then run the bot using the new command, in this case:

npm run start:collect

Or without creating the new script you can run directly:

BOTFUEL_APP_TOKEN=<the BOTFUEL_APP_TOKEN> BOTFUEL_APP_ID=<the BOTFUEL_APP_ID> BOTFUEL_APP_KEY=<the BOTFUEL_APP_KEY> node_modules/.bin/botfuel-collect-run collect-config

Format

Input file format

The input .csv file must have the following format:

The out .json file will have the following format:

[
  ...
  {
    "raw_sentence": "Utterance",
    "correct_intent": "Correct intent label",
    "sentence_after_in_middleware": "Utterance after in middlewares",
    "sentence_after_spellchecking": "Utterance after spellchecking",
    "classification_results": <CLASSIFICATION_RESULTS_LIST>,
    "sentence_after_out_middleware": "Utterance after out middlewares",
    "bot_messages": <BOT_MESSAGES_LIST>
  },
  ...
]